For years, researchers have worried that infections during pregnancy might affect a child’s developing brain.

The idea is biologically plausible. In animal studies, activating the mother’s immune system during pregnancy can alter how the offspring’s brain develops. And several human studies have found that children whose mothers had infections, including flu, were somewhat more likely to be diagnosed with conditions such as ADHD or autism.

A new study of more than 2.7 million births in South Korea and Japan tested that link in two different ways.

The first way found a small increased risk.

The second way, comparing siblings within the same family, found almost none.

Researchers used 2.7 million births in South Korea and Japan

Researchers in South Korea used national health insurance records covering nearly every birth in South Korea between 2010 and 2017, plus a separate Japanese database used to check whether the results would replicate.

  • South Korea: 2 564 435 mother-child pairs. About 36 000 mothers, 1.4%, had a recorded flu diagnosis during pregnancy.
  • Japan: 181 681 mother-child pairs. About 11 000 mothers, 6.2%, had a recorded flu diagnosis.

Children were followed until the end of 2023. The researchers looked for any later diagnosis of a neuropsychiatric condition, including ADHD, autism, obsessive-compulsive disorder, intellectual disability, and severe conditions such as schizophrenia, bipolar disorder and severe depression.

Across the population, flu in pregnancy went with slightly more diagnoses

The first analysis compared children of mothers who had flu with children of mothers who didn’t, across the whole population.

The researchers used statistical weighting to make the two groups similar in measured characteristics such as the mother’s age, income, region, number of previous children and health before pregnancy.

The result: children exposed to maternal flu had about a 9% higher risk of a neuropsychiatric diagnosis in South Korea, and about 10% higher in Japan.

Most of that came from ADHD. In South Korea, the risk of an ADHD diagnosis was about 13% higher. In Japan, about 26% higher.

Taken alone, that would look like modest evidence that flu during pregnancy affects children’s brain development.

Then the researchers changed the comparison.

They focused on families where one child was born after a pregnancy with flu and a sibling was born after a pregnancy without it. That gave about 33 000 children from such families in South Korea and about 7 800 in Japan.

In these within-family comparisons, the association largely vanished:

  • South Korea: essentially no difference in risk (hazard ratio 0.98, where 1.00 means no difference).
  • Japan: a 4% higher risk, too small and uncertain to be statistically meaningful.

Neither result was statistically significant.

The authors conclude that the population-level link should be interpreted cautiously, and that factors shared within families may account for it.

Siblings share what researchers can’t measure

This is the most interesting part of the study, and it applies far beyond flu.

When researchers compare children from different families, they can adjust for the factors they can measure: age, income, region and so on. But families differ in countless ways that don’t appear in health records.

Siblings share many of those hidden factors automatically.

Brothers and sisters share much of their genetic background. They usually grow up in the same home, with the same parents, the same habits and often the same doctors.

The study’s authors highlight two shared factors in particular:

  • Genetics. If some genetic tendencies relate both to how mothers respond to infection and to children’s neuropsychiatric risk, comparing unrelated families can create a link that isn’t caused by the infection.
  • Health care habits. Some families visit doctors more often. That makes it more likely that a mother’s flu gets diagnosed and recorded, and also more likely that a child’s ADHD gets diagnosed. Both show up in the data together, without one causing the other.

Comparing siblings is like comparing two plants grown in the same garden rather than two plants from different gardens. If the difference disappears within the same garden, the garden may have been the real story.

What the study doesn’t show

The authors are careful not to overstate their result.

Sibling studies have their own limits. They remove factors that siblings share, but not those that differ between pregnancies, such as changes in the mother’s health, circumstances or habits over time.

The sibling groups were much smaller. The researchers point out that the loss of statistical significance may partly reflect wider uncertainty rather than a true drop in risk, and should not be taken as definitive proof of familial confounding. That matters especially in Japan, where the sibling sample was small and the ADHD estimate stayed elevated, though not statistically significant.

There was one stray signal. In the South Korean sibling analysis, flu was linked to a higher risk of intellectual disability. That finding did not appear in Japan.

Only recorded flu was counted. Mothers who had flu without seeing a doctor would have been counted as unexposed.

And a null result is not proof of safety. The study does not show that flu during pregnancy is harmless. Influenza carries real risks for pregnant women and their babies, and the study does not change established advice on preventing and managing flu in pregnancy.

This is a peer-reviewed observational study, published in JAMA Network Open on October 8, 2026.

Alarming associations can depend on who is compared

Alarming headlines about pregnancy often rest on population-level associations: a slightly higher risk among people exposed to something.

This study shows how much those associations can depend on who is being compared with whom. The same data produced a modest risk when comparing unrelated families, and essentially none when comparing siblings.

For readers, the practical lesson is about interpretation. When a study links an exposure to a child’s outcome, it’s worth asking: could something about the family explain both?

Sometimes the most important variable in a study isn’t the exposure, but the family it happened in.